Method

Backdoor adjustment

Choosing an adjustment set that blocks backdoor paths.

6 cases from 6 companies apply Backdoor adjustment. 2 report results in production.

Typical tasks: Measure impact without an A/B test, Decide who gets an offer, Get experiment results faster. Companies: T-Bank, Sber, Skoltech, Walmart, DoorDash and 1 more. Often combined with: Causal inference (general), Double machine learning, Uplift modeling, CATE / HTE estimation, Bootstrap and 9 more.